Speech recognition identifies linguistic semantic content from speech. For example, speech recognition corresponds to a process of identifying a word or a word string through an input of speech waveforms and extracting semantics associated with the identified word or word string. Different techniques may be used on a speech waveform to identify words or word strings. In an exemplary embodiment, the different speech recognition techniques may be classified into five processes including speech analysis, phoneme recognition, word recognition, sentence interpretation, and semantic extraction. However, performing all of the different techniques on a speech waveform is undesirably complex. Therefore, conventional speech recognition methods generally include at least one of the processes of speech analysis, phoneme recognition, and word recognition.
Previously, the hardware necessary to implement a human-machine speech recognition interface has undesirably limited the use of speech recognition. However, with the development of large-scale integrated circuits, a speech recognition device and a voice synthesis device can be realized on an integrated circuit having a size of several millimeters in width and length, and thus allows a speech input/output device to be more easily implemented.
In an exemplary embodiment, a speech input/output device may be used to receive a bank balance inquiry, a stock price inquiry, an application of communication sales, a credit card inquiry, a hotel or flight seat reservation, etc. via phone. However, a common problem associated with general inquiries using natural language based speech question and answer (Q/A) services, such as S−Voi** or Si−*, is that there are too many questions that are unable to be understood. Further, when the Q/A services are implemented using cloud computing techniques, an increase in response delays with respect to a natural language query introduces undesirable network interruptions. When network interruptions occur, it is not possible for the device to perform additional speech recognition Q/A services.
The above information is presented as background information only to assist with an understanding of the present disclosure. No determination has been made, and no assertion is made, as to whether any of the above might be applicable as prior art with regard to the present disclosure.